I don’t quite recall the exact moment when this thought occurred to me. I found it interesting enough to share it with you – therefore this blogpost. It is about the way we – humans – do computation.

Let me give you an example, calculate 3+5.

The answer is 8.

Now how did you get there. I am sure you knew the answer already. You looked at two familier numbers 3 and 5. And you knew, you remembered the answer is 8. You did not calculate, you knew the answer and you just gave it. Maybe this was too simple calculation.

Lets take tougher example, calculate 34857 x 77173.

The answer is 2690019261.

Obviously you did not know this answer, so did we genuinly calculated this time ?

Again NO! While solving such a complex computation we break it in a simpler steps. Like first we will multiply the unit digit of second number with unit’s digit of first number, write the unit’s digit of resulting number and carry forward the carry. Then multiply unit’s digit of second number with 10’s digit of first number, add the earlier carry and so on. Finally we do vertical summation and get the answer. So here also, what we are doing is that we are just replacing the numbers, because we already know the answers of simpler calculations. So the point is once we understand the process of basic mathematical operations like addition, multiplication, subtraction and division (during our early school days), we never really compute. We just follow  procedure and replace  numbers (which in this case act nothing more than arbitrary symbols) with other numbers that’s all!

On the other hand machines (computers, calculators) don’t use these shortcut techniques. They faithfully do actual computaion every single time, however easy or difficult the calculation maybe. (Or even if you repeat the question/numeric operation) Poor machines 😦 Now this presents us a different way of designing machines who do smart work. Can we build a machine where such so called easy computations are stored or hardwired. When complex computation is required, these machines would be able to break up the problem into smaller chunks and again use hardwired answers. (I think building such machine should be possible with current technology, afterall its just a different computer not a Time-Machine. I mean, with software we can always do this, the point is can this call smart work be hardwired too) The question is how would this computer perform w.r.t its present avatar (ahem-ahem)? Now I don’t want to get into this debate. The entire purpose of this blog was to introduce you to this idea. And now that you are aware (enlightened), you can think on this offline.